On 2012-08-02 18:37, Alex Williamson wrote: > On Thu, 2012-08-02 at 11:02 +0200, Jan Kiszka wrote: >> On 2012-08-01 20:09, Alex Williamson wrote: >>> On Wed, 2012-08-01 at 09:13 +0200, Jan Kiszka wrote: >>>> On 2012-08-01 07:18, Alex Williamson wrote: >>> >>>>> +#define VFIO_IOMMU_UNMAP_DMA _IO(VFIO_TYPE, VFIO_BASE + 14) >>>>> + >>>>> +#endif /* VFIO_H */ >>>>> >>>> >>>> Please patch update-linux-headers.sh and let it do its work instead >>>> (separate patches). >>> >>> Unfortunately I missed updating the Kbuild file in the kernel, so vfio.h >>> doesn't get installed yet. I'll include that in my next pull request, >>> but maybe in the meantime I should temporarily put this header back in >>> hw/ unless you have a better idea. Thanks, >> >> Well, if it's going to be fixed very soon, we can merge it like this. >> But you should already be able provide a patch for the update script so >> that the next one running it against a fixed kernel will not >> accidentally drop the vfio header again. > > What's the reason for the rm -fr; mkdir -p in the update script? I > don't see how we'd be unintentionally generating cruft in these > directories if we were to just overwrite files instead of removing the > directory and repopulating it. I'll get the kernel fixed soon, but I'm > tempted to add vfio and drop the expunge of the directory at the same > time. Thanks, Well, chances to pile up cruft are low, granted. On the other hand, this policy enforces proper upstream header exporting as we see... Jan
Attachment:
signature.asc
Description: OpenPGP digital signature
